Commentary _____________________________________________________________
Value of MR in Definition of the Neuropathology of Cerebral Palsy in Vivo
Joseph J. Volpe 1
Selective Neuronal Necrosis
Cerebral palsy refers to a clinical syndrome of
nonprogressive motor deficits (usually spastic
weakness, uncommonly choreoathetosis, rarely
ataxia) of central origin with onset from early
infancy. In this issue of the Journal, Truwit et al
describe the MR findings in 40 patients with the
clinical diagnosis of cerebral palsy (1). The purposes of their retrospective study were to define
the anatomic substrates of cerebral palsy in the
living patient and, from the definitions of the
topography of the neuropathology and the analysis of clinical data, to suggest etiology and timing
of the brain injury. The authors place particular
emphasis on their conclusion that cerebral palsy
is usually not related to perinatal factors. The
work is relevant to the role of hypoxic-ischemic
brain injury in the genesis of cerebral palsy, the
means of diagnosis of hypoxic-ischemic brain
injury in the neonatal period, the spectrum of the
neuropathology associated with cerebral palsy
and the role of MR in the later definition of this
neuropathology, and the timing of etiologic factors. These issues are discussed in sequence.
Selective neuronal necrosis, a common expression of injury secondary to hypoxic-ischemic insult in the term infant, is characterized by neuronal destruction in a typical distribution (see ref.
2 for review). Cerebral cortical neurons, particularly those of the CA 1 region of hippocampus
(Sommer's sector), are especially vulnerable. (In
the premature infant, neuronal injury occurs principally in inferior olivary nuclei , ventral pons, and
subiculum of hippocampus, although periventricular white matter injury tends to be the predominant hypoxic-ischemic lesion in such infants.)
The pathophysiology of the typical neuronal injury of the term infant, ie, in hippocampus and
neocortex, is related predominantly to the regional distribution of excitatory synapses mediated by the excitotoxic amino acid glutamate
(see refs. 3-6 for reviews). Hypoxic-ischemic insults lead to a combination of increased release
from nerve endings and decreased uptake by
presynaptic neurons and by astrocytes and, consequently, to the toxic accumulation of extracellular glutamate. Glutamate-induced injury occurs
by rapid and delayed mechanisms. The latter may
be the more important, is calcium-mediated, and
probably generates ultimately toxic free radicals.
Current studies suggest that therapy with either
glutamate receptor blockers, calcium channel
blockers, or free radical scavengers could be of
major benefit. Regional metabolic factors related
to energy metabolism, calcium homeostasis, and
free radical scavenging capabilities, may play
contributory pathogenetic roles in the determination of the topography of selective neuronal
injury, but more data are needed on these issues.
Regional circulatory factors also may play a contributory pathogenetic role, since the cortical neuronal injury tends to be worse in arterial border
zones and end zones in the depths of sulci and
Hypoxic-Ischemic Brain Injury and Cerebral
Palsy
Hypoxic-ischemic brain injury occurring either
in the prenatal, perinatal, or early postnatal periods may result in the motor deficits categorized
as cerebral palsy (20). This brain injury is considered best in the context of five basic neuropathologic states, ie, selective neuronal necrosis, status marmoratus of basal ganglia and thalamus,
parasagittal cerebral injury, periventricular leukomalacia, and focal and multifocal ischemic
brain necrosis. Although these disorders often
overlap in a given patient, the nature of the insult
and the gestational age of the infant tend to cause
one lesion to be dominant.
1
Neurologist-in-Chief and Bronson Crothers Professor of Neurology , Department of Neurology , Children's Hospital and Harvard Medica l School, 300
Longwood Avenue , Boston, MA 02115. Address reprint requests to J . J . Volpe.
Index terms: Cerebral palsy ; Commentaries
AJNR 13:79-83, Jan/ Feb 1992 0195-6108/ 92/ 1301 -0079 © American Soc iet y of Neuroradiology
79
80
in the parasagittal regions of cerebral cortex.
The subsequent motor correlate of selective neuronal injury is usually quadriparesis, spastic or
hypotonic.
Status fvfarmoratus of Basal Ganglia and
Thalamus
Status marmoratus of basal ganglia and thalamus, an unusual form of selective injury, involves
neuronal injury, gliosis, and hypermyelination in
caudate, putamen, globus pallidus, and thalamus
(2) . The pathophysiology is related principally to
two major factors: the transient maturation-dependent dense concentration of glutamate receptors in basal ganglia of perinatal human brain (7)
and the vulnerability of neurons undergoing active differentiation in the perinatal period. Circulatory factors may contribute. The subsequent
motor correlate is choreoathetosis.
Parasagittal Cerebra/Injury
Parasagittal cerebral injury, the principal ischemic lesion of the term infant, is characterized
by cerebral cortical and subcortical injury in a
characteristic superomedial distribution, bilateral,
with the posterior cerebrum affected more than
anterior (2). Pathophysiology is based on the
presence of characteristic arterial border zones
and end zones in the parasagittal regions. These
zones are particularly vulnerable to a fall in cerebral blood flow, which is especially likely to occur
in the newborn with systemic hypotension during
and after perinatal asphyxia because of an accompanying pressure-passive cerebral circulation
(8). The latter lack of autoregulation occurs in
the asphyxiated newborn because of several interrelated factors. The subsequent motor correlate is spastic quadriparesis, with the proximal
limbs affected more than distal and the upper
extremities more than lower extremities.
Periventricular Leukomalacia
Periventricular leukomalacia, the principal ischemic lesion of the premature infant, is characterized by necrosis of periventricular white matter, dorsal and lateral to the external angle of the
lateral ventricle (2) . Although more common in
the premature infant, periventricular leukomalacia is by no means rare in the term infant subjected to ischemic insult. The pathophysiology is
related to the occurrence of periventricular arte-
AJNR : 13, January /February 1992
rial border zones and end zones, a pressurepassive state of the cerebral circulation, the relatively limited vasodilatory capability of the blood
vessels in peri ventricular white matter, relatively
active anaerobic glycolysis in periventricular
white matter, and the intrinsic vulnerability of
actively differentiating periventricular glial cells
(9). The increasing recognition of more diffuse
injury to periventricular white matter in small
premature infants (10, 11), out of the bounds of
presumed arterial border zones and end zones,
emphasizes the possibility that the intrinsic metabolic properties of periventricular white matter
may be as important as the vascular factors. The
subsequent motor correlate of periventricular leukomalacia is spastic diplegia, a kind of spastic
quadriparesis in which lower extremities are affected much more than upper extremities.
Focal and /Vfultifocallschemic Brain Necrosis
Focal and multifocal ischemic brain necrosis is
characterized by injury to all cellular elements (ie,
an infarction) in a vascular distribution (2). Single
or multiple vessels, arteries or veins, may be
involved. The distribution of the middle cerebral
artery is the single most common vascular territory affected. The pathophysiology is multifactorial. Obstructions of vessels by vascular maldevelopment, vasculopathy, embolus and thrombus
have been documented. By mechanisms not yet
clear, apparent generalized circulatory insufficiency in the perinatal period also may be followed by focal or multifocal cerebral ischemic
lesions. The subsequent motor correlates consist
of spastic hemiparesis or spastic quadriparesis.
Diagnosis of Hypoxic-Ischemic Brain Injury in
the Neonatal Period
The diagnosis of the five lesions described
above in the living infant has been based primarily
on brain imaging, ie, cranial ultrasonography
(US), computed tomography (CT), and recently,
magnetic resonance (MR), and on neurophysiologic techniques. Selective neuronal necrosis currently is assessed most commonly by neurophysiologic techniques, especially electroencephalography (EEG), but also brain stem auditory evoked
responses and visual evoked responses. Such
EEG patterns as burst-suppression or persistent
and pronounced voltage suppression are indicative of serious cortical neuronal injury (2). However, less severe degrees of neuronal injury are
AJNR: 13, January/ February 1992
not defined clearly by neurophysiologic techniques. Moreover, in our experience, brain imaging modalities have not been highly useful in the
neonatal period in identification of cortical neuronal injury, except in its most severe form.
However, experience with MR in this setting still
is relatively limited. (The later cortical atrophy,
which correlates with neuronal loss and gliosis,
of course is detected readily by CT and MR.)
Injury to basal ganglia and thalamus is detected
best by MR. Thus, Keeney et al (12) showed that
MR detected hypoxic-ischemic basal ganglia injury in five infants, whereas US identified abnormality in four of the five infants and CT in only
one of the three studied by this modality.
Parasagittal cerebral injury has been identified
most readily by radionuclide brain scanning (2,
13) or positron emission tomography (2, 14), only
uncommonly by CT (15), and never, in our experience, with US. However, the recent demonstration of neonatal parasagittal cerebral injury
by MR (12, 16) suggests that this lesion now
may be detectable with this modality and, of
course, with superior localization to radionuclide
scanning.
Periventricular leukomalacia is identified most
readily in the newborn by cranial US (see ref. 9
for review). Bilateral periventricular echogenic
areas, especially in the peritrigonal regions, in the
first week, correlate with necrosis accompanied
by vascular congestion and/ or hemorrhage;
echolucent foci, after 1 to 3 weeks in the same
distribution, correlate with formation of periventricular cysts; ventricular enlargement, often with
disappearance of cysts after 3 months or more,
correlates with diminished myelin deposition and
collapse of cysts with gliosis. MR also is particularly useful in the identification of periventricular
white matter injury in the neonatal period (12),
and in subsequent months and years, the appearance of increased signal on T2-weighted images in the periventricular white matter is a particularly prominent indicator of old periventricular
leukomalacia (17).
Focal and multifocal ischemic brain necrosis
generally is effectively identified by CT (except
in the first day or two after the insult), although
the recent data of Keeney et al suggest that MR
may prove to be superior (12). Acute focal cerebral ischemic lesions also are identifiable as echogenic lesions by US, although smaller and more
peripherally placed infarcts, eg, in posterior parietal regions, may be missed by US.
81
Spectrum of Neuropathology Associated with
Cerebral Palsy as Defined by Later MR
The study of Truwit et al (1) provides important
information regarding the spectrum of neuropathology associated with cerebral palsy in patients
between 1 month and 41 years of age and the
value of MR in defining that spectrum. Their first
observation of particular interest is that MR detected an abnormality of brain in 93 % of their
patients with cerebral palsy (1). This is a remarkable yield for any modality in such a heterogeneous group of patients.
The neuropathology defined in the premature
infants consisted of a diminution in periventricular
white matter in every patient (1). Moreover, the
findings were consistently suggestive of periventricular leukomalacia. Thus, the topography of
the white matter involvement usually was peritrigonal and/ or accompanied by prolongation of T2
relaxation. Interestingly, three patients also had
small brain stems, perhaps in part because of the
pontine and olivary neuronal necrosis referred to
earlier. In nine of the 11 infants, the perinatal
course was characterized by clinical circumstances suggestive of fetal or neonatal hypoxicischemic insults, usually neonatal. No developmental abnormalities of brain were noted in the
premature infants. Thus, these data suggest that
cerebral palsy in the premature infant is related
to periventricular myelinoclastic disease, occurring most often in the postnatal period, and is
rarely related to underlying developmental brain
abnormality.
The spectrum of neuropathology defined in the
term infants with cerebral palsy differed clearly
from that in the premature infants. Thus, 31 % of
the 29 term infants exhibited evidence for neuronal migrational disturbance (1). The largest proportion (59%) had an apparently heterogeneous
collection of abnormalities that included particularly diminished cerebral myelin, with cortical
thinning and/or cerebral infarction and/or hydranencephaly and/ or multicystic encephalomacia
and/or basal ganglia and thalamic lesions. Only
10% had no definable abnormality. The relatively
high proportion of disturbances of neuronal migration included particularly focal cerebral cortical disturbances with an "abnormally thick" cortical mantle, characterized by Truwit et a! (1) as
"polymicrogyria" (eight of nine cases), and a single case of schizencephaly. Of the 59% with the
heterogeneous collection of abnormalities, apparent cerebral cortical atrophy (thinning of cerebral
cortex) was present in three cases and abnormal
82
signal intensities in basal ganglia and thalamus in
six cases. We consider these two groups the
probable correlates of selective cerebral cortical
neuronal injury and basal ganglia/thalamus injury
as described earlier. The abnormalities in white
matter are more difficult to characterize. Thus,
19 term infants exhibited diminution of myelin,
but in only eight cases were long T2 signals
observed, and in only nine cases was the diminution of myelin peritrigonal (1). These findings
suggest that only approximately 50% of the infants with myelin abnormality had periventricular
leukomalacia. The remaining approximately 50%
of cases with diminished myelin did not have
features highly suggestive of periventricular leukomalacia and appeared to require other explanations for the myelin diminution . Such explanations would include a developmental defect of
myelin formation, perhaps analogous to that recorded in neuropathologic studies by Richardson
and coworkers (18), or a secondary disturbance
of myelin deposition because of impaired axonal
formation . The impaired axonal formation could
be present either as a consequence of neuronal
migrational disturbance (six of the 19 infants with
diminution of myelin had evidence for neuronal
migrational defect as defined above), an intrinsic
derangement of neuronal outgrowth, ie, a socalled organizational event of brain development
(2), or prior neuronal injury with subsequently
stunted axonal development.
Timing of Lesions Responsible for Cerebral
Palsy in Term Infants
The timing of the abnormalities in the term
infants with cerebral palsy relates to the nature
of the pathologies. Thus, the neuronal migrational
disturbances must originate during the peak time
period for migrational events in the human cerebral cortex, ie, 3 to 5 months of gestation. The
nature of the focal cerebral cortical disturbance
described by Truwit et al (1) as polymicrogyria
was characterized by a thick cortical mantle, a
finding which is compatible with pachygyria as
well as with polymicrogyria . If the latter, the
disturbance would be expected to originate in
approximately the fifth month of gestation; if the
former, the disturbance would be expected to
originate in the fourth month of gestation (2). The
timing of the apparent selective neuronal injury
involving cerebral cortex or basal ganglia and
thalamus is unclear. In several patients, perinatal
hypoxic-ischemic insults appear to have been
AJNR: 13, January/ February 1992
responsible, although the retrospective nature of
the study precludes more refined definition of
timing. The timing of the disturbance of cerebral
myelin that appears to be related to periventricular leukomalacia is unclear, ie, an ischemic insult
either in the third trimester of pregnancy or in
the perinatal period could produce the abnormality. A developmental disturbance of neuronal
development with secondary impairment of myelination could have its onset from the latter
months of gestation to approximately the first
year of postnatal life. Similarly, a developmental
disturbance of cerebral myelination per se presumably would have its onset around the time of
birth.
Truwit et al considered that 5 of the 29 term
infants (17 %) had MR and perinatal historical
data suggestive of "perinatal" hypoxic-ischemic
injury (1). Two additional infants (7%) had data
suggestive of both prenatal and perinatal injury.
Despite the problems of attempting to draw conclusions about timing from a retrospective study,
this value (ie, 17%-24%) is similar to that obtained in several large-scale epidemiologic studies
of patients with cerebral palsy. Thus, studies in
recent years from Sweden, Finland, the United
States, Australia, and Ireland suggest a perinatal
origin of cerebral palsy in approximately 12%24% of cases who were term infants (19-26).
Although Truwit et al (1) emphasized that "only
24 % exhibited perinatal brain injury, it should be
recognized that, because of the relatively high
prevalence of cerebral palsy, this 24% represents
a large absolute number of infants. Moreover,
although current methods of fetal monitoring
have not led to a marked decrease in the number
of infants with cerebral palsy secondary to perinatal events, it remains entirely reasonable to
suspect that perinatal causes of cerebral palsy
ultimately should be preventable to a large degree. Clearly, newer methods of fetal monitoring
and definition of exact timing and mechanisms
of perinatal brain injury are needed.
Conclusions
This most interesting study by Truwit et al (1)
shows that cerebral palsy in the premature infant
is related primarily to periventricular white matter
injury, perhaps occurring primarily in the neonatal
period. Cerebral palsy in the term infant has a
more heterogeneous basis; ischemic injury to
cerebral cortex, basal ganglia and thalamus, and
to myelin accounts for some, but by no means
AJNR: 13, January /February 1992
all, cases. Timing is perinatal in approximately
17%-24% of such cases. Strikingly, cerebral
cortical dysgenesis secondary to disordered neuronal migration appears to account for nearly
one-third of cerebral palsy in term infants. MR is
of major value in identification of the anatomic
substrates for cerebral palsy.
83
imaging of the central nervous system. II. Lesions associated with
hypoxic-ischemic encephalopathy. Pediatrics 199 1;87 :431-438
13. Volpe JJ , Pasternak JF. Parasagittal cerebral injury in neonatal
hypox ic-ischemic encephalopathy: clinical and neuroradiological features. J Pediatr 1977;91 :472-477
14. Volpe JJ, Herscovitch P, Perlman JM, et al. Positron emi ssion
tomography in the asphyxiated term newborn : parasagitta l impairment of cerebral blood flow. A nn Neuro/1 985; 17:287-296
15. Pasternak JF. Parasagittal infarction in neonatal asphyxia. A nn Neural
References
1. Truwit CL, Barkovich AJ, Koch TK , Ferriera DM. Cerebral palsy: MR
findings in 40 patients. AJNR 1992; 13:67-78
2. Volpe JJ. Neurology of the newborn. 2nd ed. Philadelphia: W. B.
Saunders, 1987.
3. Zivin JA, Choi DW. Stroke therapy . Sci Am 1991;265:56-63
4. Hattori H, Wasterlain CG. Excitatory amino acids in the developing
brain: ontogeny, plasticity, and excitotoxicity. Pediatr Neural
1990;6:219-229
5. MacDonald JW, Johnston MV. Physiological and pathophysiological
roles of excitatory amino acids during central nervous system development. Brain Res Rev 1990;15:41-70
6. Benveniste H. The excitotoxin hypothesis in relation to cerebral
ischemia. Cerebrovasc Brain Metab Rev 1991 ;3:213-245
7. Barks JD, Silverstein FS, Sims K, Greenmayre JT, Johnston MV.
Glutamate recognition sites in human fetal brain. Neurosci Lett
1988;84:131-136
8. Pryds 0 , Greisen G, Lou H, Friis-Hansen B. Vasoparalysis associated
with brain damage in asphyxiated term infants. J Pediatr
1990;117:119-125
9. Volpe JJ. Current concepts of brain injury in the premature infant.
AJR 1989; 153:243-251
10. Paneth N, Rudelli R, Monte W, Rodriguez E, Pinto J , Kairam R. White
matter necrosis in very low birth weight infants: neuropathologic and
ultrasonographic findings in infants surviving six days or longer. J
Pediatr 1990;116:975- 984
11 . Dambska M , Laure-Kamionowska M, Schmidt-Sidor B. Early and late
neuropathological changes in perinatal white matter damage. J Child
Neuro/1989;4:291-298
12. Keeney SE, Adcock EW, McArdle CB. Prospective observations of
100 high-risk neonates by high-field (1.5 tesla) magnetic resonance
1987;21 :202-203
16. Pasternak JF, Predey TA , Mikhael MA. Neonatal asphyx ia: vulnerability of basal ganglia, thalamus and brainstem. Pediatr Neural
1991;7:147-149
17. Flodmark 0, Lupton B, Li D , et al. MR imaging of periventricular
leukomalacia in childhood . AJNR 1989;10:111-118
18. Chattha AS , Richardson EP Jr. Cerebral wh ite matter hy poplasia.
Arch Neuro/1977;34:137-148
19. Hagberg B, Hagberg G. The changing panorama of infantile hydrocephalus and cerebral palsy over forty years: a Swedish survey . Brain
Dev 1989; 11 :283-290
20. Hagberg B, Hagberg G, Olow I, von Wendt L. The changing panorama
of cerebral palsy in Sweden. V. The birth year period 1979-82. Acta
Paediatr Scand 1989; 78:283-290
2 1. Hagberg B, Hagberg G, Zetterstrom R. Decreasing perinatal m ortality:
increase
in
cerebral
palsy
morbidity?
Acta
Paediatr
Scand
1989;78:664-670
22. Riikonen R, Raumavirta S, Sinivuori E, Seppala T. Changing pattern
of cerebral palsy in the Southwest region of Fi nland. Acta Paediatr
Scand 1989; 78:581-587
23. Nelson KB, Ellenberg JH. Antecedents of cerebral palsy. N Eng/ J
Med 1986;3 15:81-86
24. Blair E, Stanley F J. Intrapartum asphyxia: a rare cause of cerebral
palsy. J Pediatr 1988; 112:515-51 9
25. Grant A, O 'Brien N, Joy MT, et al. Cerebral palsy among children
born during the Dublin randomised trial of intrapartum m onitoring.
Lancet 1989; 1:1 233-1 235
26. MacDonald D, Grant A, Sheridan-Pereira M , et al. The Dublin randomized controlled trial of intrapartum fetal heart rate monitoring. Am J
Obstet Gyneco/1985; 152:524-539